Taking evidentialism to be the view that you are justified in fully believing that P only if you have adequate evidence that P, Adler argues that evidentialism follows from the very nature of full belief. His conclusion is that full belief is essentially an attitude subject to the demands of evidentialism, so that it is actually inconsistent to believe that P and that also that you do not have adequate evidence that P!
Question: What has to be true of full belief if Adler is right?
We are given:
The Inconsistency Claim suggests that, in fully believing that P, what you believe implies you have adequate evidence that P. If so, the content of your full belief cannot be simply that P, because the proposition that P by itself typically does not imply you have adequate evidence that P.
If Adler is right, what you believe in fully believing that P must imply not only that P and but also that you have adequate evidence that P.
Let us assume for a while that the Conjunction Claim is correct. This does not imply that the second conjunct about adequate evidence is explicitly represented in your mind, in some sort of “full belief box.” If you form beliefs by putting representations in your belief box, then in order fully to believe that P it may be sufficient to put an explicit representation with the content that P into your full-belief box without also having to put into your full-belief box any additional explicit representation of the content that you have adequate evidence that P. That further content might be implicit in placing a representation with the content that P into your full-belief box. In fact, the content probably must be implicit in your placing a representation into your full-belief box, since in this view, you cannot put a representation that P into your full-belief box without automatically also representing the content that you have adequate evidence that P.
If this is right, I agree with Adler that there must be a functionalist account of it. To say that full belief involves that sort of complex content has to be to say you act as if it does--you think as you would think if full belief involved that sort of complex content. That is, the extra content must be explainable in terms of the way full belief functions in your psychology, at least when all goes well. This means that (a) you are not fully to believe something unless you can also believe that you have adequate evidence for it and (b) given that you believe something you can take yourself to have adequate evidence for it, where principles (a) and (b) follow from and define the nature of full belief, distinguishing it from other related attitudes like guessing, assuming, and tentatively accepting something for the sake of argument.
We are assuming that the content of a full belief that P is at least as strong as the conjunction “P and I have adequate evidence that P.” Question: is this whole conjunction itself fully believed? If so, it follows that you must also believe you have adequate evidence for it and so for infinitely many further iterations!
Again, this would not imply that there are infinitely many distinct representations in your full-belief box. All of the infinite content except for the content that P might be implicit in your having come fully to believe that P. But it would imply the so-called BB Principle for full belief:
What about the corresponding KK Principle for knowledge?
The KK Principle might be defended by appeal to a couple of other principles. One is:
Since (let us assume) your knowing that P implies you have adequate evidence that P, the principle that Full Belief Is Acceptance as Known would account for why it is inconsistent to believe that P while also believing you do not have adequate evidence that P.
The other relevant principle is:
These two principles, that Full Belief Is Acceptance as Known and that Knowing Is Knowing It All, together imply the KK principle: you know that P only if you know that you know that P and so only if you know that you know that you know that P, etc., where much of what you know is implicit rather than explicitly represented.
Suppose that the following principle is correct.
Then, it would follow from the content of what you believe that you have adequate evidence for what you believe and, redundantly, adequate evidence that you have adequate evidence for it, etc.
Similarly, if the following principle is correct:
then it would follow from what you know in the first place that you know you know it, know you know you know it, and so forth.
Someone might object that content cannot be self-referential because, if it is possible for content to be self-referential, paradoxes arise. For example, you could have the thought, “this very thought is not true,” and someone might argue that having such a thought is impossible because if it were to exist then it would be both true and not true.
But prohibiting direct self-reference is not enough to eliminate semantic paradoxes and in any event there are many cases in which self-reference is quite intelligible (Kripke, 1975).
Furthermore, as I will now explain, there are various other phenomena of philosophical interest that are best treated as involving self-referential content, namely, speaker meaning, common knowledge, intentions, and perceptual experiences. Looking at these other cases sheds some light on full belief.
Grice’s (1956) initial account of speaker meaning appealed to a self-referential intention.
Grice added, “This seems to involve a reflexive paradox, but it does not really do so.” Later, as various complications were noted, Grice (1969) and Schiffer (1972) replaced the self-referential analysis with ones involving a series of intentions, with later intentions about the earlier intentions. This let to issues about the existence of the potentially infinite regress of intentions required, issues that could have been avoided by staying with self-referential formulations.
Lewis (1969) and Schiffer (1972) offered an analysis of what it is for it to be common knowledge that P among a group of people. Their basic idea was
Again this sort of analysis led to worries about how many such iterations an ordinary person can handle. A better analysis of this sort appeals to self-referential knowledge with a relatively simple clause that implies all the iterations.
Intentions are best treated as self-referential. The point is clearest for positive intentions, which are to be distinguished from negative intentions and conditional intentions. Roughly speaking, positive intentions are ones you envision as leading to the intended result, as when you intend to go to today’s Faculty Meeting.
Negative intentions, like intending not to go to today’s Faculty Meeting are different. You do not think of your intention as responsible for your not going to the meeting. It would be enough not to intend to go to the meeting.
Conditional intentions are an intermediate case. They are envisioned as leading to the desired result only on the condition that a certain condition is met. You intend to go to the meeting if your class ends before 3 pm.
I claim:
Negative and conditional intentions are somewhat different.
The principle that Positive Intentions Are Self-Referential explains an important aspect of practical reasoning. In order to reason to a positive intention to do something, you must be able to conclude as part of that reasoning that your conclusion will lead to the result intended. If you are convinced that your decision cannot affect what will happen, then you cannot reason practically about it and so cannot form the positive intention in question (Harman, 1976).
The principle that Positive Intentions as Self-Referential specifies an important difference between intentions and desires or hopes, which need not be self-referential in that way. The principle is also relevant to explaining what it is to do something intentionally (Searle 1983, Harman 1976).
Even though your positive intentions refer to themselves, this reference can be implicit in the way you use the representation you form in forming the intention. So, it is possible, for example, for young children to form intentions even if they have no explicit concept of intentions or of self-reference.
The principle that Positive Intentions Are Self Referential principle does not apply to negative intentions, like intending not to go to the Faculty Meeting, because negative intentions do not work by getting the intender to bring about the intended state of affairs. This raises the question why you even bother to form such a negative intention, and the answer seems to be that you do so in order to settle the matter as to whether you is going to the meeting. And that answer suggests a more general principle that applies to all intentions--positive, negative, and conditional:
In the case of positive intentions, you can settle it that what you intend will happen only by intending to bring that about. But that is not required for negative and conditional intentions.
Perceptual experience has an intentional content that represents the environment in a certain way. It represents how the environment is “from here.” Furthermore, the content of that representation is plausibly self-referential (Searle 1983, Harman 1990).
It is often said that the difference between intention and belief has to do with the “direction of fit.” I suggest that one such difference between these types of mental state can be expressed in the following two principles.
I now want to compare the Self-referential Guarantee in Belief Content with the principle of Self-reference and Adequate Evidence discussed earlier.
Let us assume with Adler that adequate evidence for the truth of P settles it that P in context. Then the principle of Self-reference and Adequate Evidence goes beyond the Self-referential Guarantee in Belief Content by requiring that what settles it that P is “adequate evidence” that P.
Adler uses the term “evidence” in a very wide sense to stand for any sort of reasons you might have, including the sorts of reasons for mathematical conclusions. The relevant sort of evidence is not limited to observational evidence.
Suppose your “evidence” has to be something that you currently believe, or are currently aware of, and consider a belief that is the result of something in the past that settles it that P, although you do not now remember what it was that settles it that P. Then, the Self-referential Guarantee in Belief Content would allow the relevant full-belief but, let us suppose, the principle of Self-reference and Adequate Evidence would not.
But consider that you do not keep track of your original reasons for your beliefs. You believe hundreds of thousand or millions of things for which you no longer remember your original reasons and for which you have not come up with new reasons beyond your belief itself.
Perhaps if you were to consider the matter, you could defend any given belief in terms of others. But the fact that noticing such a defense would give you adequate evidence that P does not mean that you have adequate evidence that P before you have noticed the defense.
So, if evidence has to be understood in the suggested way (as something you believe or are currently aware of), the principle of Self-reference and Adequate Evidence disallows most of your beliefs simply because you have not kept track of your reasons for them. Assuming with Adler that most of your beliefs are not to be disallowed in this way, it follows that, either the principle of Self-reference and Adequate Evidence has to be rejected as implausible in favor of the Self-referential Guarantee in Belief Content, or your evidence that P can include something e in the past that settles it that P even if you no longer remember e, in which case the principle of Self-reference and Adequate Evidence coincides with the Self-referential Guarantee in Belief Content. So, we might as well appeal directly to the Self-referential Guarantee in Belief Content rather than to the principle of Self-reference and Adequate Evidence.
You are epistemically justified in believing that P only if, either (a) the mere fact that you believe that P is enough to make you justified in believing that P, or (b) your belief that P is justified by other considerations beyond the fact that you believe that P.
Present time internalist theories suppose that such justification must appeal only to your presently existing psychological states or events, possibly including unconscious states or events, but not including either partly external considerations such as reliability in context or historical considerations of which you are no longer aware.
According to present time internalism, internal duplicates with the same beliefs are equally epistemically justified in their beliefs. Reliabilist and historical entitlement theories imply that such duplicates might not be equally justified in their beliefs, an implication that has generally been regarded as so counter-intuitive as to constitute a conclusive objection to such views (Cohen, 1984; Foley, 1985; Ginet, 1985; Pollock, 1984). Recently, philosophers like McDowell (1955) and Williamson (2000) have argued for the counter-intuitive result on the grounds that it is the only way to escape skepticism. But they are simply wrong to think this counter-intuitive move is the only way to escape skepticism.
Present time internalist theories differ as to which if any of your beliefs might be epistemically justified simply by virtue of the fact that they are believed. Coherence theories suppose that no beliefs have this sort of foundational status and special foundationalist theories suppose that only some of your beliefs can have that status. General foundationalist theories suppose that all of your current beliefs have that status in the sense that they are all at least prima facie justified simply in virtue of the fact that they are believed. The familiar idea is that you start reasoning from where you are, using everything that you currently believe, making changes to eliminate conflicts in your beliefs, to add answers to questions in which you are interested, etc., with the possibly unattainable goal of reaching what Rawls calls “reflective equilibrium” (Rawls, 1971, referring to Goodman, 1955).
Adler rejects any sort of foundationalism, including general foundationalism, as incompatible with evidentialism. However, I am not sure what his favored alternative is. At times he appears to think your belief that P might be justified by virtue of the fact that there exists a defense of that belief even if the belief is not in any way based on that defense. He also seems to appeal to historical considerations such as that your background beliefs have in the past received confirmation in various ways. I won’t try to work out his exact position on these matters.
Instead, I want to consider go back to comparing the two principles:
My suggestion is that the Self-referential Guarantee in Belief Content is the right principle and that, when you believe that P, the something that settles it that P might well be something that isn’t part of your current evidence, so it does not provide adequate evidence for your belief that P.
Finally, I observe that Adler’s main argument for evidentialism seems better suited to be an argument for the Self-referential Guarantee in Belief Content rather than for the principle of Self-reference and Adequate Evidence. His argument is that his understanding of evidentialism explains why you cannot randomly believe something such as that the number of stars in the universe is even. That is you cannot randomly do that while realizing that you are doing it. His explanation for your inability is that it is incoherent for you in full awareness fully to believe both that the number of stars is even and that you do not have adequate evidence that the number of stars is even.
The alternative explanation that I favor would be that it is incoherent for you in full awareness to believe both that the number of stars is even and that you do not have that belief because of something that settles it that the number of stars is even.
One reason for favoring the alternative explanation is that Adler’s account implies the incoherence of religious faith in full awareness, whereas the alternative does not, because you might believe that your belief in God is (in part) explained by something that settles it that God exists. Adler acknowledges that his treatment of religious faith as incoherent is surprising and not obviously correct. One of the main goals of his book is to argue from his theory against the initially attractive idea that religious faith is quite coherent.
Although the Self-referential Guarantee in Belief Content can be used to defend the coherence of religious faith, there is no similar way to use it to defend the coherence in full awareness of believing something at random. It is not coherent to suppose both that your belief is explained by something that guarantees its truth and also that it is a random belief. So the Self-referential Guarantee in Belief Content explains Adler’s data without yielding his counter-intuitive conclusion about religious faith.
I conclude that Adler is “sort of right” about what is involved in full belief. But what he is right about is not enough to defend the strong evidentialism he favors.